1,000 KiloBytes, but commonly refers to 1,024 KibiBytes which is actually called a MebiByte. Originally, the "kilo"/"mega"/etc prefixes were used for 2^(10*x) amounts in computer jargon, while they usually mean 10^(3*x). This was not a problem, since computer and non-computer jargon generally didn't overlap. When computers began pervading all aspects of life, some people deemed this duality confusing and proposed to introduce new "kibi"/"mebi"/etc prefixes for the base-2 factors and to use the "kilo"/"mega"/etc prefixes for traditional base-10 factors. However, this distinction has not been adopted in practice, so it's not always clear what exact capacity is meant by a particular mention of MegaByte.

In speech, it's often cut down to "meg".

This is a very common unit. The majority of sets of data dealt with in contemporary computing range from somewhere between half a MegaByte to about 50. 400-600 MegaBytes is a common size for MultiMedia data, owing to the fact that that's roughly the capacity of CDs.

See also KiloByte/KibiByte, MebiByte, GigaByte/GibiByte, TeraByte.